(dissenting).
I dissent because I think that the court has overturned the district judge's factual findings in an unwarranted fashion. The court’s reversal of the decision below is primarily based on its dispute with the district judge over the proper reading of the facts.
I.
We remanded to the district court for the following reasons:
In an extremely close case such as this, with a record that admits of conflicting interpretations, a court of appeals is ill-equipped to undertake its own de novo assessment of the facts against the proper standard. On remand, the district court should inquire whether and when a reasonable person in Streifel and Quinn’s position would have believed that he was actually in police custody and being constrained to a degree associated with formal arrest (rather than simply undergoing a brief period of detention at the scene while the police sought by means of a moderate number of questions to determine his identity and to obtain information confirming or dispelling their suspicions). ... On remand, the court may, but is not required to, take additional evidence, and may reopen and redetermine any issue covered in this appeal as, in its discretion, it sees fit.
United States v. Streifel, 781 F.2d 953, 962 (1st Cir.1986) (footnote omitted). We fur*162ther stated that, after making this factual finding, the court could order suppressed statements elicited in violation of Miranda as well as other evidence that it deemed illegally obtained. Id.
Determining whether a Terry stop exhibits restraints equivalent to those of a formal arrest involves questions both of law and of fact. At the extremes, there are cases which may be decided on almost purely legal grounds. At one end of the spectrum are cases where police suspicions are based on factors so general that they are insufficient to allow even the minimally intrusive Terry stop. Terry v. Ohio, 392 U.S. 1, 21 n. 18, 88 S.Ct. 1868, 1880 n. 18, 20 L.Ed.2d 889 (1968) (specific information required for police to make Terry stop). See, e.g., Brown v. Texas, 443 U.S. 47, 99 S.Ct. 2637, 61 L.Ed.2d 357 (1979) (suspicion based merely on individual’s presence in a high-crime area insufficient). At the other end of the spectrum are cases where the “trappings of a formal arrest” are so great that detention cannot be justified under “reasonable suspicion” but require probable cause. See, e.g., Dunaway v. New York, 442 U.S. 200, 99 S.Ct. 2248, 60 L.Ed.2d 824 (1979). Most cases, however, are not governed by “bright line” rules but must be decided on the basis of an evaluation of the circumstances. See, e.g., United States v. Sharpe, 470 U.S. 675,105 S.Ct. 1568, 1575, 84 L.Ed.2d 605 (1985).
In our prior opinion, we clearly stated that this case could only be decided through such an evaluation of the circumstances. Moreover, because it was an “extremely close case,” we declared ourselves “ill-equipped” to make such a determination. Streifel, 781 F.2d at 962. In other words, we held that this case turned on the facts and that the legal parameters governing the limits of Terry stops were not dispositive in one direction or the other.
The district judge cited two kinds of factors to support his view that the situation became equivalent to an arrest when Holmes and Steadman returned. First, following our suggestion that he review the facts in the light of Berkemer v. McCarty, 468 U.S. 420,104 S.Ct. 3138, 82 L.Ed.2d 317 (1984), he noted that the traffic stops the Court discussed in that case, like Terry stops, are presumptively temporary and brief. An individual who is subject to such a stop expects that he will answer a few brief questions and be on his way. See id. at 437-38, 104 S.Ct. at 3148-49. The return of the two officers and the blocking of the driveway, according to the district judge, made the situation more like a prolonged, pressured, station-house interrogation. See id.
Secondly, the combined circumstances produced an atmosphere which, the district judge found, “can most reasonably be described as police-dominated.” See Berkemer, 468 U.S. at 438-39, 104 S.Ct. at 3149-3150. Among the factors the judge listed were: (1) the presence of five police officers and a large, trained dog; (2) the continued retention of the car registration; (3) the keeping apart of defendants by the officers, indicating a strategy to produce confession;4 (4) the nonpublic nature of the interrogation.5 The district judge concluded: “Having been blocked in at this remote spot, Defendants had reason to feel at the mercy of the police and to feel that they would be detained indefinitely until they satisfied the police.”
*163The court’s attack on the district judge’s factual analysis focuses on each element individually in order to conclude that no single one could have tipped the balance in favor of an arrest. The court holds that neither the addition of two police officers, nor the blocking of the driveway, nor the duration of detention, nor the characteristics of the area in which the stop occurred, was sufficient to support the district court’s holding. The district judge, however, did not rely on any single factor as legally dispositive, but looked at the cumulative impact of the various elements of restraint. While he did regard the blocking of the driveway as a “determining factor,” he recognized that such blocking “does not necessarily elevate the Terry stop into an arrest.” His decision, rather, was based on the “total factual context” thereby following our directive to make “fact-specific assessments and inquiries” of the situation as a whole, an inquiry which we declared ourselves unable to undertake.
I can understand this aspect of the court’s opinion, therefore, in one of two ways. First, the court may feel that the district judge’s factual findings were, as such, clearly erroneous. I see no basis in the record for such a view. Second, the court may view one or more of the elements considered by the district judge as legally irrelevant to the evaluation of the situation as a whole. I find no basis in the law for this position. As the court notes, citing United States v. Sharpe, 470 U.S. 675, 105 S.Ct. 1568, 1575, 84 L.Ed.2d 605 (1985), the Supreme Court has stressed that “common sense and ordinary human experience must govern over rigid criteria” in making the determinations in these matters. The district judge thus appropriately considered the total atmosphere of the situation in arriving at his decision.
Moreover, the court’s opinion becomes confusing in its discussion of the duration of the stop. The court correctly begins its analysis with Sharpe, a case which rejected any “hard-and-fast, time limit” beyond which a stop becomes an arrest. Sharpe, 105 S.Ct. at 1575. The court then goes on to make its own assessment of the facts, overruling the analysis of the district judge. The court states that Steadman had “not only reasonable grounds, he had very strong grounds, approaching probable cause” and it would therefore have been “unreasonable to have sent Quinn and Streifel on their way after a few perfunctory questions.” I question whether this new finding of suspicion “approaching probable cause” is consistent with our first decision, let alone with the factual finding below. It is clear, though, that it renders the court’s opinion internally inconsistent. For after justifying a prolonged detention (perhaps one “approaching” arrest?) and declaring the unreasonableness of release, the court goes on to restate that the detention was not overly restraining: “Except for being directed to await Steadman’s arrival, Quinn and Streifel were not told to stay nor advised that they were in custody, nor were they physically threatened or held.”
The court cannot have it both ways. Terry stops involve some restriction on individuals’ freedom of movement. Yet, it cannot be factually true that there was both a prolonged detention, whose highly constraining nature was justified on grounds “approaching probable cause,” and a detention that was minimally intrusive with the defendants not constrained from leaving. The court’s version of the facts is not internally consistent and is certainly not compelling enough to warrant overriding the findings of the district judge.
II.
A. Suppression of Evidence under the Fifth Amendment
Miranda established a per se rule. Statements elicited from persons in custody who have not been warned of their rights are inadmissible. Because I would uphold the district court’s finding of custody, I would suppress all statements made after the return of Officers Steadman and Holmes.
B. Suppression of Evidence under the Fourth Amendment
The analysis for suppression under the fourth amendment is somewhat different. *164Evidence may satisfy the requirement that it be obtained “voluntarily” under the fifth amendment and yet still be inadmissible under the fourth amendment as a product of an illegal detention. Brown v. Illinois, 422 U.S. 590, 601-02, 95 S.Ct. 2254, 2260-61, 45 L.Ed.2d 416 (1975). Even Miranda warnings are insufficient, without more, to purge the taint of an illegal detention. Id. Rather, a court must analyze whether the “casual connection” has been broken. Id. at 603, 95 S.Ct. at 2261.
Having ruled that the detention here became an arrest without probable cause at the time of the return of Holmes and Steadman, the district judge considered whether the evidence obtained after that time still bore the taint of that illegality. Once an arrest has occurred, he noted, police officers may not exploit the situation to build the probable cause they lacked in the first instance. The Supreme Court has often disapproved of such investigatory arrests. See Florida v. Royer, 460 U.S. 491, 503, 103 S.Ct. 1319, 1327, 75 L.Ed.2d 229 (1983); Dunaway v. New York, 442 U.S. at 215-16, 99 S.Ct. at 2258-59; Brown v. Illinois, 422 U.S. 590, 605, 95 S.Ct. 2254, 2262, 45 L.Ed.2d 416 (1975). The district judge therefore ruled that the sniff test was tainted by the illegal detention and that its results were inadmissible.
The district judge then turned to Quinn’s purported consent to the car search. He relied on Brown v. Illinois, where the Court discussed a confession obtained as a product of an illegal detention. The Court stated:
The question whether a confession is the product of a free will under Wong Sun must be answered on the facts of each case. No single fact is dispositive. The workings of the human mind are too complex, and the possibilities of misconduct too diverse____
Brown, 422 U.S. at 603, 95 S.Ct. at 2261. The district judge stated that, given the atmosphere of police domination, Quinn’s “consent” could not satisfy the “casual chain” test under the fourth amendment. In addition to the elements of the situation I have listed supra at pages 162-63, he noted that the officers had told Quinn that they could get a warrant if he did not consent to the search. This threat of a warrant cast further doubt on the freely given nature of Quinn’s consent. See United States v. Ocheltree, 622 F.2d 992, 994 (9th Cir.1980) (consent not voluntary when given under the threat of a warrant); cf. Florida v. Royer, 460 U.S. at 503, 103 S.Ct. at 1327 (discussing impact on consent of police plan to obtain warrant).
In reversing the district judge, the court does not address the problems under Wong Sun and Brown v. Illinois. It holds that the sniff test was therefore justified, and that the results of that test gave the officers probable cause to search the car without a warrant.
Because I would uphold the district court’s finding of an illegal arrest, I would find that the sniff test was a product of that illegality and was an illegal attempt to supply the missing probable cause. The nature of Quinn’s consent remains a difficult question. This issue should be decided in line with the Supreme Court’s distinction in Brown v. Illinois, 422 U.S. 590, 95 S.Ct. 2254, 45 L.Ed.2d 416, between consent that would be “voluntary” and consent that would be strong enough to break the “causal chain.” Quinn twice consented to the searches and “represented himself to the officers to be a lawyer.” Yet he had earlier refused his consent to the search and agreed only after the sniff test and the threat of a warrant. The district judge noted the resulting sense of the “futility” of a refusal, thus vitiating the voluntariness of the consent. On such an “extremely close” question, involving the degree of consent based on a “record that admits of conflicting interpretations,” I would follow our prior decision and defer to the district judge’s assessment. I would, therefore, uphold the suppression of the warrantless searches of Quinn’s car, as well as the fruits of the search conducted with the warrant secured with illegally obtained evidence.
. Cf. Miranda v. Arizona, 384 U.S. 436, 448, 86 S.Ct. 1602, 1614, 16 L.Ed.2d 694 (1966) ("the modern practice of in-custody interrogation is psychologically rather than physically oriented").
. Cf. Berkemer v. McCarty, 468 U.S. at 438, 104 S.Ct. at 3149:
Perhaps most importantly, the typical traffic stop is public, at least to some degree. Passersby on foot or in other cars, witness the interaction of officer and motorist. This exposure to public view both reduces the ability of an unscrupulous policeman to use illegitimate means to elicit self-incriminating statements and diminishes the motorist’s fear that, if he does not cooperate, he will be subjected to abuse. The fact that the detained motorist typically is confronted by only one or at most two policemen further mutes his sense of vulnerability. In short, the atmosphere surrounding an ordinary traffic stop is substantially less "police dominated” than that surrounding the kinds of interrogation at issue in Miranda [and its progeny].